ChatSpatial

概述

ChatSpatial 是一个专为空间转录组学分析设计的 MCP 服务器，通过模式强制编排替代了即兴的 LLM 代码生成。它提供了 20 个经过模式验证的 MCP 工具，编排了 15 个分析类别中的 65 种空间转录组学方法。该项目提供了一个稳定的自然语言界面，用户可以通过简单的提示词处理空间数据，而不是编写复杂的脚本，使空间转录组学工作流程在不同会话和客户端之间更具可重复性。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• load_data

  Load spatial transcriptomics data from files like h5ad

  试试:Load /absolute/path/to/spatial_data.h5ad试试:Load my spatial data from the desktop
• show_tissue_structure

  Visualize tissue structure of loaded spatial data

  试试:Show me the tissue structure试试:Display spatial layout of the data
• spatial_domain_identification

  Identify spatial domains using methods like SpaGCN, STAGATE

  试试:Find spatial domains in the tissue试试:Identify tissue regions using SpaGCN
• deconvolution

  Perform cell type deconvolution using methods like Cell2location

  试试:Deconvolve cell types in my data试试:Run Cell2location to estimate cell proportions
• cell_cell_communication

  Analyze cell-cell communication with LIANA+ or CellPhoneDB

  试试:Find communicating cells in the tissue试试:Run CellPhoneDB to identify ligand-receptor pairs
• cell_type_annotation

  Annotate cell types using methods like scANVI or SingleR

  试试:Annotate cell types in my data试试:Use scType to identify cell types
• differential_expression

  Find differentially expressed genes across conditions

  试试:Find genes that differ between tissue regions试试:Run differential expression analysis
• trajectory_inference

  Perform trajectory inference with CellRank or Palantir

  试试:Find cell trajectories in my data试试:Run CellRank to identify differentiation paths
• spatial_statistics

  Calculate spatial statistics like Moran's I or co-occurrence

  试试:Calculate Moran's I for gene expression试试:Find spatially variable genes
• enrichment_analysis

  Perform gene set enrichment analysis like GSEA

  试试:Run GSEA on my differentially expressed genes试试:Find enriched pathways in my data
• visualization

  Create spatial plots, embedding plots, and expression overlays

  试试:Create a spatial plot of gene expression试试:Show UMAP embedding colored by cell type
• multi_sample_integration

  Integrate multiple samples using Harmony or BBKNN

  试试:Integrate my 10 spatial samples together试试:Run Harmony to batch correct my data

可对比工具

安装

pip install chatspatial

docker pull ghcr.io/cafferychen777/chatspatial:v1.2.9

{
  "mcpServers": {
    "chatspatial": {
      "command": "python",
      "args": ["-m", "chatspatial", "run"]
    }
  }
}

FAQ

ChatSpatial 支持哪些空间转录组学平台？

ChatSpatial 支持 10x Visium、Xenium、Slide-seq v2、MERFISH 和 seqFISH 平台。

如何加载数据？

只需提供类似'加载 /absolute/path/to/spatial_data.h5ad 并显示组织结构'这样的自然语言提示即可，ChatSpatial 会内部处理数据加载。

ChatSpatial 对比